EVALUATING MORPHOLOGICAL DIFFERENCES BETWEEN TWO MICROBIAL MAT
MORPHOTYPES FROM LITTLE AMBERGRIS CAY AS A TOOL FOR STUDYING THE
GEOLOGICAL RECORD
by
Sydney Riemer
Undergraduate Thesis
Advisor: Dr. Maya Gomes
Department of Earth and Planetary Sciences, Johns Hopkins University
April 20, 2018
2
Abstract
Fossilized microbial mats are important records of early life on Earth. The changing
macroscale morphology of these lithified microbial mats through time has been interpreted as
either the result of biological evolution or changing environmental conditions. Microbial mats
from active carbonate platforms serve as modern analogues and possible precursors to lithified
microbial mats, thereby allowing us to test hypotheses about microbial mat formation, growth,
and morphology. In this study, we apply geological methods used to study the sedimentary
record of ancient microbial mats to two modern microbial mat morphotypesthe polygonal and
flat matsfrom Little Ambergris Cay, Turks and Caicos Islands. We investigate the upper
pigmented layers of these mats, and the buried layers that represent previous generations of
surface microbial mat communities. Using scanning electron microscopy, we obtain point counts
of mat taxa and widths of mat cyanobacteria. We find that filament widths between the
polygonal and flat mats are indistinguishable, and that the filament widths and microfaunal
abundances of the lower layers of the active mats are similar to the old mats. In addition, we
document differences in microfaunal abundances and filament widths between the surface layers
and old counterparts of the polygonal and flat mats, indicating environmental changes through
time. These results imply that mat morphologies are influenced by environmental conditions
rather than microbial communities, and are consistent with findings from gene sequencing
studies. Overall, the results of this study demonstrate that distinct mat morphologies occur not
because of biological diversity, but environmental factors and have implications for studying
paleoenvironmental change in sedimentary records of lithified microbial mats.
3
1. Introduction
1.1 Microbial Mats and Stromatolites
The geological record is a useful archive of the co-evolution of life and the environment,
yet can be difficult to interpret due to the post-depositional alteration of rocks and sediments by
tectonic activity and diagenesis, which distort paleoenvironmental records. Therefore, when
studying the geological record, it is helpful to look at modern systems that may serve as
analogues to the geologic feature being studied. One such geologic feature that has benefited
from studies of modern analogues are stromatolites, often interpreted to be lithified microbial
mats. Stromatolites are present in some Archean rocks, become ubiquitous in shallow marine
Proterozoic sediments, and then decline at the start of the Phanerozoic (Peters et al., 2017).
Stromatolites are formally defined as attached, lithified sedimentary growth structures,
accretionary away from a point or limited surface of initiation. They are also distinctively
layered, with individual laminae representing incremental growth of the microbial mat over time
as sediment is accreted (Grotzinger and Knoll, 1999), although they can also grow by abiotic
processes (Grotzinger and Rothman, 1996). The two commonly accepted mechanisms for
microbially-mediated stromatolite growth and accretion are: (1) the trapping and binding of
sediments in microbial mats and (2) the precipitation of carbonates and other inorganic materials
through the activity of microbial mats (Grotzinger and Knoll, 1999). Stromatolites are significant
because they can be archives of ancient organisms, ecosystems, and environments on the early
Earth and throughout geologic time. However, both biological and environmental processes
influence stromatolite forms found in the geological record. Yet, diagenetic alteration of rocks
makes it difficult to identify the taxonomic and metabolic diversity of microorganisms found in
stromatolites (Knoll et al., 2013). This leads to the question of whether variations in stromatolite
form are due to the evolution of mat-building microorganisms or changing environmental
conditions. Thus, there is the potential to uncover important information on biological evolution
and global environmental change from morphological signatures in stromatolites. Here, I present
a study of modern microbial mats as potential stromatolite precursors in order to evaluate the
biological and environmental factors that influence microbial matand possibly also
stromatoliteformation, growth, and morphology.
Microbial mats are multilayered structures of microorganisms across all three domains of
life with diverse lifestyles and metabolisms. Microbial mats also contain the metabolic
byproducts of these microorganisms as well as trapped sediments, mineral precipitates, and
allochthonous material (Des Marais, 2003). They are found all over the Earth in environments
ranging from Antarctic lakes to the deep sea. Microbial mats typically adhere to a pattern of
vertical stratification. The two layers found in most mats exposed to light are a green layer and a
pink layer. The top green layer is typically on the mm- to cm- scale and is host to
photoautotrophs and heterotrophs. Of these, the most important for mat-building are
photosynthetic cyanobacteria (Knoll et al., 2013). In the pink layer below the green layer, light
penetrates but oxygen does not because it is consumed in the top green layer. Therefore, this
layer is host to anoxygenic photoautotrophs such as purple sulfur bacteria that give this layer its
pink color and use H
2
S rather than H
2
O for photosynthesis, as well as anaerobic heterotrophs and
chemoautotrophs. Microbial mats accrete upward over time, and material and biomass in the
upper layers becomes substrates for heterotrophs in the lower layers. The lower layers of the
mats are typically brown in color and consist of degraded and decomposed biomass of past
upper-mat layers. The material in the lower brown layers is the material that has the potential to
4
enter the geologic record. This vertical stratification pattern is important to consider because
modern microbial mats serve as analogs to those preserved in the geologic record, and therefore
we can compare what diversity we see in the surface layers of modern microbial mats to the
lower layers that contain previous generations of the mats to learn what aspects of biodiversity
are preserved in lithified microbial mats. Thus, microbial mats in carbonate environments offer
insights into the physical, chemical, and biological processes that occur as these diverse
ecosystems go through various stages of degradation and decay during burial, and how these
various processes affect mat structure and stromatolite formation.
1.2 What aspects of Earth history do lithified microbial mats record?
Lithified microbial mats occur as a variety of morphological forms and fabrics that
change both temporally throughout geologic time and spatially with depositional environment.
Common stromatolite morphologies include stratiform, domal, columnar, and coniform (Knoll et
al., 2013). It is inferred that stromatolite macrostructure is ultimately influenced by the shape and
relief of the lamina, and therefore it is the microstructural characteristics of stromatolites that are
studied to determine how different stromatolite morphologies originate. Microstructural
characteristics that are often studied include cyanobacterial filament orientation, size and shape,
and inorganic-precipitate fabric (Knoll et al., 2013; Grotzinger and Knoll, 1999). Though
stromatolites have been studied for over a century, it is still unclear what processes are most
important in influencing microbial mat and stromatolite morphology. Elucidating this can help
researchers determine what components of stromatolite morphology represent information about
biological diversity and evolution, global and local environmental change, or both.
Researchers have come at the stromatolite morphogenesis problem from a variety of angles using
different methods and techniques. Grotzinger and Knoll (1999) use numerical simulations to
argue that changing stromatolite morphology over time is a result of changes in the saturation
state of seawater carbonate. The changing carbonate saturation state of seawater then determines
whether stromatolites will form by trapping and binding of sediments or in-situ precipitation of
carbonate, two stromatolite formation mechanisms that result in different morphologies
(Grotzinger and Knoll, 1999).
Another study made observations of Proterozoic stromatolites from the Angmaat
Formation in Canada, which records intertidal to supratidal carbonate deposition (Knoll et al.,
2013). They found stromatolites of varying morphologies there, and by analyzing which
microfabrics are correlated with different microfossil assemblages, they determined that
stromatolite microfabrics and therefore macro-morphology vary depending on whether the mat is
composed of mostly coccoid bacteria, vertically oriented cyanobacteria filaments, or both
horizontally and vertically oriented filaments. As a result, their conclusion was that
microstructural fabrics of stromatolites, and therefore also morphology, are influenced by the
microbial populations. Filament orientation is also one of the characteristics most influential to
the structure. Additionally, they showed that stromatolites with the most diverse microfossil
populations occur in depositional environments where they are more frequently submerged, and
that those with less diverse populations occur in environments where there is more subaerial
exposure. Therefore, to some extent, environmental conditions also affect stromatolite structure.
The modern microbial mats in this study occur in intertidal environments and can be used to test
the effect of water depth on mat diversity, though not to the same extent as Knoll et al. (2013)
because of the less extreme change in water depth.
5
With the development of genomics and gene sequencing, researchers can now obtain a
more accurate estimate of the identity and diversity of microbial populations in microbial mats.
Because many microorganisms, especially cyanobacteria, have morphological features that may
seem to distinguish species, but which are homoplastic, gene sequencing is crucial for the
taxonomic classification of microbial mat organisms (Trembath-Reichert et al., 2016). A recent
study from Little Ambergris Cay, Turks and Caicos Islands, the same field site from this study,
used 16S rRNA gene sequencing and compared the cyanobacterial and non-cyanobacterial
populations of two microbial mat morphotypes to test the hypothesis of whether microbial
population affects mat morphology (Trembath-Reichert et al., 2016). Notably, the mat
morphotypes they chose for their study are the polygonal and flat mats, the same morphotypes
used in this study. They found that while cyanobacteria are responsible for the structural
components of both mat types, the cyanobacterial populations of the two mats are essentially the
same, indicating that it is not the taxonomic classification and diversity of the primary mat-
builders that influences mat morphology. The study did find that there are significant differences
in the non-cyanobacterial populations of the two mat types, and that the mat type found in the
interior of the lagoon where there is less hydrodynamic variability (polygonal mat) is more
diverse than the mat type found closer to the main tidal channel that has more hydrodynamic
activity and disturbance happening around it (flat mat). Lastly, Trembath-Reichert et al. (2016)
hypothesized that the polygonal mat morphology might develop from an initially flat mat
morphology, where diversity increases over time when there is little environmental disturbance
to the mat. This hypothesis was based on previous work from Wanless et al. (1988) that
categorized the mat morphotypes based on the presence of cyanobacteria with different
colonization strategies. The flat mat had a “Schizothrix” morphology that was described as a
rapid colonizer whereas the polygonal mat had a “Scytonema” morphology that was slower
growing and colonized areas that were previously covered by “Schizothrix.” Trembath-Reichert
et al. (2016) combined this finding with conclusions from other studies on modern stromatolites
that showed that hydrodynamics and sedimentation were important factors controlling
morphology (Andres and Reid, 2006). Therefore, they concluded that mat morphology is
influenced by environmental conditions, in this case environmental disturbances such as
sedimentation and erosion due to storms or proximity to tidal channels.
1.3. Objectives and Hypotheses
The purpose of this study is to use modern microbial mats with distinct morphologies to
evaluate differences between the microbial populations of the two mats types, and what aspects
of these differences might enter the geological record and yield information about either
biological evolution or environmental change. In order to do so, I investigate microfaunal
assemblages in two microbial mat morphotypes from Little Ambergris Cay, Turks and Caicos
Islands, British Overseas Territories to test hypotheses about biological and environmental
influences on microbial mat morphology, as well as the potential preservation of these mat
communities in the geological record. Little Ambergris Cay is a carbonate platform where
sediment fluxes are limited (Gomes et al., in revision). Therefore, the microbial mats on Little
Ambergris Cay are not lithified and also do not participate in much sediment trapping or binding.
This allows us to observe how the physical, chemical, and biological processes at play in the
system affect mat morphology before lithification. It also enables us to view the mat
microstructures without the distorting influence of lithification or sediment trapping. This makes
6
Little Ambergris Cay mats particularly useful for testing hypotheses on the processes of
degradation and decay that influence what stromatolites in Precambrian carbonates record.
Microbial mats on Little Ambergris Cay, like Precambrian stromatolites, occur as a
variety of morphotypes. The three observed morphotypes are classified based on texture and
morphology, and vary with elevation (Stein et al., 2016). The mat type occurring at the lowest
elevation is the permanently submerged extracellular polymeric substance (EPS) coated flat mat.
The polygonal mats with tufts of cyanobacterial sheaths are tidally submerged, and the blister
mats are only submerged during storms (Fig. 1; Stein et al., 2016). This study considers the EPS-
coated flat mats and tufted polygonal mats because they are the two most common types of mats
on the Caicos platform (Trembath-Reichert et al., 2016) and can be used to test hypotheses about
mat morphology and water depth.
In this study, one flat mat and one polygonal mat were sampled, and sections were taken
of the green and pink layers of both mats and imaged using scanning electron microscopy
(SEM). Samples from the lower brown layers that represent buried surface layers were also taken
and imaged, because while it is important to look at the upper, live mat layers, the lower old mat
could also be representative of what might enter the geological record. By comparing the old
layers to the upper layers of the active mat, we can see what information is lost upon decay and
degradation of the mat and what preservational biases there may be. Imaging software
(JMicroVision 1.2.7) was used to do point-counts of mat features and organisms and make
measurements of bacteria filament widths. Though it was shown using 16S rRNA gene
sequencing that the polygonal mats are taxonomically more diverse than the flat mats,
(Trembath-Reichert et al., 2016), researchers studying lithified microbial mats in the rock record
do not have the ability to use gene sequencing to determine which microbial species are present
and how diverse the population is. Instead, many studies of lithified microbial mats and
microbial fossils in general use microfossil morphology as a way to differentiate between taxa
such as coccoidal vs. filamentous bacteria, and filament width as an indicator of species,
specifically when it comes to cyanobacteria (Mackey et al., 2015; Knoll et al., 2013; Boal and
Ng, 2010; Schopf, 1993; Shixing and Huineng, 1992; Barghoorn and Tyler, 1965). Therefore,
measuring filament width rather than using genomics in the modern mat samples across the two
morphotypes can allow us to test if their cyanobacterial populations have similar morphological
characteristics, while obtaining a result that may be more useful and have more application for
geological studies of ancient microbial mats. Statistical analyses were performed in Python on
the point-count and filament width data to test hypotheses about how mat macrostructure
provides information about microbial communities and/or environmental conditions and how this
information might be recorded in the geological record.
A few different hypotheses were tested using point-count data of mat microorganisms
and filament width data of probable cyanobacteria. First, based on the morphological diversity of
organisms, we determine what morphological differences there are overall between the two mats.
Next, we determine if the cyanobacteria population of the polygonal mat is indistinguishable
from that of the flat mat based on the widths of the cyanobacteria filaments. Finally, we
determine if the old, buried polygonal and old flat mats have any similarity to either the active
polygonal and flat mats or each other, and the significance of this for the flat to polygonal mat
succession hypothesis as well as paleoenvironmental analysis. Overall, we aim to understand the
potential of morphological signatures in extant microbial mats to provide information about
ancient microbial mat ecosystems and/or environmental conditions.
7
2. Methods
2.1 Study site
Little Ambergris Cay is an uninhabited island on the Caicos Platform with an area of
about 6 km x 1.6 km (Fig. 2). It is at an elevation of 1-1.5 m above sea level and is framed by a
bedrock rim of oolitic grainstone (Gomes et al., in revision; Orzechowski et al., 2016; Stein et
al., 2016). The interior basin of the island is dominated by microbial mats and mangroves (Stein
et al., 2016). Water depth varies tidally by up to 50 cm, although this variability is less in the
island interior. The microbial mats in the shallow bays and channels of the island interior vary in
morphology with elevation and water depth (Gomes et al., in revision). The polygonal mats
occur as discrete quasi-polygons while the flat mats are laterally continuous (Fig. 3).
The polygonal and flat mats used for this study were sampled in July 2016 and August
2017. The polygonal mat, sample CC17-4, was collected on August 4, 2017 (Fig. 3 and 4). The
top pigmented layers of CC17-4 contain a 5 mm thick surface layer of dark green tufts, followed
by a 1 mm lighter green-white layer that contains calcium carbonate grains, a 2 mm green layer
with filaments, and lastly, a 2-4 mm pink layer. Beneath these layers are alternating layers of
brown, filamentous layers and ooid layers (Fig. 4). The flat mat, sample CC3, was collected on
July 4, 2016 (Fig. 3 and 5). The top pigmented layers of CC3 are a 2-3 mm thick, grainy lime
green layer followed by a 1 mm pink layer (Fig. 5). The unpigmented layers are brown,
filamentous and granular layers. The mats were stored in plastic bags and refrigerated until
sectioning for microscopy. In this study, the top pigmented layers and a section of older material
from the unpigmented layers were studied for both mat morphotypes. Representative SEM
images of all four mat samples are shown in Figure 6.
2.2 Scanning electron microscopy
Scanning electron microscopy (SEM) was chosen as the imaging technique because it
allows us to view the structure of the mat and the relative positions of its components, which is
useful for obtaining structural information such as filament orientation. SEM sample processing
requires sectioning and fixing the mat samples in order to obtain the best quality images. For the
polygonal mat, one ~3 cm thick section of the green and pink pigmented layers was taken for
imaging as well as one ~2 cm thick section of a brown filamentous layer starting from ~4 cm
deep in the mat. This layer was chosen because there were visible vertical tufts, and because it
occurs deep enough in the mat that we can observe an entirely different generation of mat
growth. This allows us to test hypotheses about microbial mat succession after storm or other
disturbance events as well as what from the active polygonal mat might be preserved in the
degraded polygonal mat that could enter the geologic record. From the flat mat, one ~2 cm thick
section was taken of the green and pink pigmented layers and one ~2 cm thick section of a tan
layer from ~4.5 cm deep in the mat was taken to again test hypotheses about preservation in the
geologic record. The unpigmented layers represent the old, buried mats that contain material that
might enter the geological record while the pigmented, active layers capture the processes that
occur before degradation, decay, and possibly lithification. Prior to sectioning, a sterilized razor
blade was used to remove all material that had come into contact with the plastic bag. Sterilized
razor blades were used for the carving and removal of the individual sections from the mats.
Fixation of the mat is necessary so that the electron beam in the SEM does not destroy the
sample and render it unavailable for further use. The fixation protocol followed in this study is
the protocol followed at the Integrated Imaging Center at Johns Hopkins University (IIC), where
8
all sample preparation and imaging took place. All four mat sections were fixed with 3.0%
formaldehyde and 1.5% glutaraldehyde in 0.1 M sodium cacodylate and 5 mM Ca
2+
, 5 mM
MgCl
2+
, 2.5% sucrose, pH 7.4, at room temperature for one hour. Samples were then washed
three times in 0.1 M cacodylate, 2.5% sucrose, pH 7.4, for fifteen minutes for each wash.
Afterwards, samples were post-fixed with Palade’s OsO
4
for one hour in the dark. 5 mL of
Palade’s 1% OsO
4
is composed of 1 mL Acetate-veronal sock, 1.25 ml 4% OsO
4
, 1 ml 0.1 N
HCl, and 1.75 ml deionized water. After osmication the mat samples become discolored and it is
difficult to tell with certainty what was the original “up” direction. Therefore, 0.10 mm stainless
steel minutien pins were embedded into the top layers of all samples before fixation in order to
preserve the original orientation. After the fixation process, samples were immersed in ethanol
and dried using a Tousimis 795 critical point dryer. After samples were dried they were mounted
on SEM stubs and sputtered with platinum in an Anatech Hummer 6.02 Sputter Coater. The
processed samples were then kept in SEM stub boxes until imaging.
All SEM imaging was done on the FEI Quanta 200 Environmental SEM at the IIC. For
each of the active and old polygonal mat sections and each of the active and old flat mat sections,
twenty sites were sampled for imaging at intervals of roughly every 1-3 mm. Images were taken
from both the green and pink layers of the active mats. Two images were taken at each site, one
at ~1000x and another at around ~3500x. Images were taken at these two magnifications in order
to reduce bias towards either large or small organisms and features. High voltage mode was used
for all images. In all subsequent sections, discussion of the polygonal mat and flat mat refer to
the combination of results and analysis from both the green and pink layers of those mats.
2.3 Image analysis
All SEM images were analyzed using the image processing software JMicroVision. For
each of the two images taken at the twenty sites from the four mat sections, 100 counts of mat
organisms and features were done using the random point counting tool in JMicroVision. Widths
were taken for cyanobacteria and bacteria filaments that were identified during point counting
using the spatial calibration and 1D measurement tools.
2.4 Statistical methods
Different statistical methods were employed using Python for analyzing the point count
data and the filament width data. In order to analyze the point count data of the various features
and species/morphologies in the mats, two statistical methods were used to help visualize and
quantify the data. First, to visualize differences between the polygonal, flat, old polygonal, and
old flat mats, principal component analysis (PCA) was applied to the data in a variety of
different ways to investigate relationships in the data. Across the four different mat samples, nine
different microorganism morphologies and features were identified during point counting. The
relative abundances of these features were then calculated. The most abundant of these features
by far was amorphous organic matter, the classification used for the decayed and featureless
organic matter in the mats. In order to explore how similar the mats are based on both the whole
sample and identifiable organisms present, PCA analysis was done on the point count relative
abundance data with and without amorphous organic matter. Additionally, a third PCA analysis
was done on the data with the removal of singletons and other non-organismal features that were
observed. Doing these multiple PCA analyses allows for the data to be analyzed in different
ways to see how the mats are similar or different in terms of their organisms present as well as
other identifiable features that may affect mat structure and morphology.
9
The second statistical method used to analyze the point count data and determine how
similar the different mat samples are to one another was non-metric multidimensional scaling.
The Bray-Curtis dissimilarity metric is commonly used in ecology to evaluate the dissimilarity
between two different sites. The Bray-Curtis dissimilarity metric was chosen for this study
because it can be used on relative abundance data from different locations, or in this case,
different microbial mat types and ages. The Bray-Curtis dissimilarity metric returns a percent
dissimilarity between two mat types and/or layers (e.g. percent similarity between polygonal and
flat mats, or percent similarity between the polygonal green layer and the polygonal pink layer)
that can be used to determine how similar the mat types and layers are relative to one another by
looking at the percent similarity, which is 1-percent dissimilarity. Because all mat types and
layers have a >83% relative abundance of amorphous organic matter, Bray-Curtis dissimilarity
analysis was carried out on the relative abundances including amorphous organic matter and
normalized without it in order to obtain a better estimate of the differences between the microbial
populations. Absolute abundances of microfauna and features counted using point counting and
analyzed by PCA and Bray-Curtis dissimilarity are in Table A1.
For the filament widths, the Welch’s t-test, or the unequal variances t-test, was used to
test the hypothesis that that the cyanobacteria filament widths in the polygonal mat are
indistinguishable from those in the flat mat. The Welch’s t-test was also used to determine if the
cyanobacteria filament widths are indistinguishable between the old mat samples and the active
mat samples in order to help determine if the old samples represent buried polygonal or flat mats.
The Welch’s t-test was used because of the differing variances and sample sizes of filament
widths from the different mat types and layers (Table A2). Because Welch’s t-test assumes
normality, the test was performed on filaments with widths from 0-5 μm, a range where the data
are approximately normal. However, Welch’s t-test is robust to non-normality (Ruxton, 2006),
allowing the test to be performed on the whole range of filament width data as well. Therefore,
Welch’s t-test was carried out on different combinations of comparisons for the mat samples, and
t-statistics and p-values were obtained to determine whether or not filament widths between the
mat samples are significantly similar. All analyses including Welch’s t-test, PCA, and Bray-
Curtis dissimilarity were performed using the Python programming language’s SciPy statistical
packages and their functions (Welch’s t-test: scipy.stats.ttest_ind; PCA:
sklearn.decomposition.PCA; Bray-Curtis dissimilarity: scipy.spatial.distance.braycurtis). See the
SciPy documentation for more information on those functions.
3. Results
3.1 Point counting
The most abundant mat morphology/feature identified during point counting across all
four mat samples was amorphous organic matter, which accounts for >83% of features observed
in the mats (Fig. 7 and 8). The polygonal mat and its green and pink layers have both the least
amorphous organic matter and the most cyanobacterial filaments. The old mats have the most
amorphous organic matter, though the flat mat has a similar relative abundance of amorphous
organic matter as the old polygonal and old flat mats (Fig. 7 and 8). The other features/taxa
identified in the mats are cyanobacterial filaments, other bacterial filaments (bacterial filaments
that aren’t obviously thick cyanobacteria filaments and therefore could be other bacterial taxa),
coccoidal bacteria, colonial coccoidal bacteria, diatoms, eukaryote feces, dinoflagellates, and
larval mollusks.
10
Removing amorphous organic matter and renormalizing the relative abundances of the
mat features and taxa makes it easier to see the abundances of identifiable microfauna (Fig. 9 and
10). Therefore, all subsequent discussion of results refer to those analyses that were done without
taking into account amorphous organic matter, unless specified. The top three most abundant
microfauna across all four mats are bacterial filaments, cyanobacterial filaments, and coccoidal
bacteria. Colonial coccoids and diatoms are also relatively significant in the polygonal, flat, and
old flat mats. Diatoms make up a significant fraction of the microfauna in the flat mat and are
also present in the polygonal mat and the old mats. Although diatoms would not be present in
Precambrian stromatolites because of their evolution later in the Phanerozoic, they are a
significant component of the Little Ambergris Cay microbial mats, and therefore were kept in the
analyses. The polygonal green layer has the most cyanobacterial filaments, which account for
~42% of that mat layer. The flat mat has the least amount of cyanobacteria filaments. The other
mat features occur as singletons or are non-microbial (Fig. 9 and 10). The polygonal, flat, and
old flat mats have seven different identifiable taxa, whereas the old polygonal mat only had six
distinct features.
PCA analyses were carried out on the relative abundance data in three permutations; (1)
with amorphous organic matter, (2) normalized without amorphous organic matter, with
singletons and non-microbial features, and (3) normalized without singletons and non-microbial
features. The removal of the amorphous organic matter and singletons is justified here because
when studying the geological record, researchers often only take into account the identifiable
morphologies. Similarly, researchers studying fossil microbial mats would be able to observe
singletons and handle them in their analyses. PCA analyses were used to visualize similarities
between the bulk mats, as well as the individual mat layers. For PCA carried out on the bulk
mats and individual layers with amorphous organic matter, the flat mat and the old polygonal
mat plot close to each other, indicating their similarity (Fig. 11a,b). On the individual-mat-layer
level, the old polygonal mat and the flat mat pink layer plot close to one another, which is
consistent with what was observed in the PCA plot for the bulk mats (Fig. 11b). When the
analysis is done without taking into account the relative abundance of amorphous organic matter,
the results are slightly different. On the bulk mat level, none of the mats plot close to one
another, though the flat mat and the old polygonal mat plot relatively close (Fig. 11c). On the
other hand, for the individual mat layers, the flat mat pink layer and the old polygonal mat still
plot close together like they did for the analysis done including amorphous organic matter (Fig
11d). Next, PCA was done on the data without singletons, and with and without amorphous
organic matter. For the bulk mat data, none of the mats plot close to each other for the analysis
done both with and without amorphous organic matter (Fig. 12a,c). However, for the individual
mat layer analyses with and without amorphous organic matter, the old polygonal mat and the
flat mat pink layer plot close to each other like they did in the analyses including the singletons
and non-microbial features (Fig. 12b,d). A new relationship emerged in the analysis without
singletons and without amorphous organic matter between the old flat mat and the polygonal mat
pink layer, which plot close to one another (Fig. 12d). Overall, the mat types and layers that
show similarity in their taxa and features based on PCA are the old polygonal mat and the flat
mat, the old polygonal mat and the flat mat pink layer, and the old flat mat and the polygonal mat
pink layer.
In order to quantify the similarity between the mat types and layers based on relative
abundance of taxa and other features obtained from point counting, the Bray-Curtis dissimilarity
metric was used. The Bray-Curtis dissimilarity analysis was done without considering
11
amorphous organic matter. Note that numerical data in Figures 13 and 14 are the percent
similarity between the mat types and layers, where percent similarity=1-percent dissimilarity and
multiplied by one hundred. For these analyses all percent similarities are >50%. For the analysis
done using relative abundances on the bulk mats, the flat and old polygonal mats have the
highest percent similarity (90.4%), a result that is consistent with PCA, and the polygonal and
old flat mats have a similar percent similarity (89.3%) (Fig. 13). The old polygonal and the old
flat mats also have a similar percent similarity, at 86.6%. The polygonal and the old polygonal
mats, and the flat and the old flat mats, have percent similarities of 76.4% and 80.5%,
respectively (Fig. 13). The least similar mats based on point count data are the polygonal and the
flat mats.
Lastly, the Bray-Curtis analysis was performed on the individual mat layers to see how
similar they are to each other and the old mat layers (Fig. 14). The comparisons with the highest
percent similarity are the polygonal pink layer and the old flat mat (92.7%), the polygonal pink
layer and the old polygonal mat (89.3%), and the flat mat pink layer and the old polygonal mat
(87.9%). The latter is a result that is consistent with PCA. The least similar layers were the
polygonal green and flat green layers, and polygonal green and flat pink layers. The flat green
layer and old flat mat, and the polygonal green layer and old polygonal mat also have relatively
low percent similarities. Another notable comparison is the polygonal pink and flat pink layers
(83.7%). All other comparisons are shown in Figure 14. The results obtained from the Bray-
Curtis dissimilarity metric are mostly consistent with the results from PCA and allow us to
quantify the relative similarities between the different mat types and their layers.
3.2 Filament widths
Across all mat types and layers, bacterial filament widths ranged from less than 1 μm to
70 μm. Thin filaments are defined here as filaments less than 10 μm, and typically ranged from
0-5 μm. Thick cyanobacterial filaments ranged from 10-70 μm. Of the polygonal, flat, old
polygonal, and old flat mats, the polygonal mat has the thickest filament widths, with an average
filament width of 7.89 μm (Table A2). The flat mat has the thinnest filament widths, with an
average filament width of 3.27 μm (Table A2). Of the green and pink layers in the polygonal and
flat mats, the polygonal green layer had the highest average filament width, and the flat green
layer had the lowest (Table A2). The number of filament widths measured, their mean, standard
deviation, and variance for each mat type and layer are reported in Table A2. A histogram of the
filament widths across all mat types and layers is shown in Figure 15. The filament widths that
were encountered at the highest frequency are those in the range from ~0-5 μm. This trend is the
same at both the individual mat and layer levels (Fig. 16-19). Because of this distribution, t-tests
were also performed on only the 0-5 μm filament width data (Table A3) in order to see if the
mats are significantly similar in that filament width range.
Using a significance level of 0.05, from Welch’s t-test across the entire filament width
range the following mat filament width comparisons for the bulk mats have significantly similar
widths: the polygonal and old polygonal mats, the polygonal and old flat mats, and the old flat
and old polygonal mats (Fig. 20). On the individual mat-layer-level, the following have
significantly similar widths: the polygonal green and polygonal pink layers, the polygonal green
and pink layers and old polygonal mat, the polygonal green and pink layers and old flat mat, the
flat green and flat pink layers, and the old polygonal and old flat mats. (Fig. 21). All other
comparisons have significantly different filament widths across their entire filament width ranges
(Fig. 20 and 21). T-statistics for the p-values in Figures 20-23 are listed in Tables A4 and A5.
12
For the Welch’s t-test for filament widths in the range 0-5 μm, all mat comparisons have
significantly similar filament widths except for the old polygonal and old flat mats, the polygonal
green and polygonal pink layers, the old flat mat and the flat green layer, and the old flat mat and
polygonal green layer (Fig. 22 and 23).
4. Discussion
A notable result from the PCA and Bray-Curtis dissimilarity analysis is that the pink
layers of the polygonal and flat mats are more similar to the old mat layers than the green layers
(Fig. 12d and 14). Although this may be surprising given that previous work has showed that
thick cyanobacterial sheaths, which tend to be more abundant in the upper green layer, have a
higher preservation potential than other mat components (Newman et al., 2017), recent work by
Gomes et al. (in revision) on a polygonal mat from Little Ambergris Cay obtained results that are
consistent with this result. They found that labile organic macromolecules such as lipids and
polysaccharides that are abundant in the green layers of the mats are quickly degraded and
depleted in the lower pink layer (Fig. 24). Those results indicate that the pink layers are already
partially degraded and so are closer to what may be preserved in the geological record.
Therefore, the findings here using microscopy are consistent with the results from previous
organic geochemistry studies that showed that the pink layers are more degraded, and in fact
almost just as degraded, as the material in the previous mats. This means that there are some
aspects of microbial diversity that are erased as the uppermost layers become lower microbial
mat layers. It is thus important to consider the mode and timing of lithification, as this likely has
a great effect on what can be preserved (Newman et al., 2017). For instance, it has been shown
that cyanobacteria can calcify and induce carbonate precipitation in microbial mats by increasing
pH through the photosynthetic uptake of inorganic carbon, thereby inducing lithification in the
upper layers where most cyanobacteria reside (Benezrara et al., 2014; Altermann et al., 2006). In
addition to this, cyanobacterial calcification and carbonate precipitation in microbial mats is also
dependent on environmental conditions such as temperature and seawater alkalinity, which can
affect where and when lithification occurs (Grotzinger and Knoll, 1999; Riding, 1992).
Therefore, though the lower mat layers may be more representative of what might enter the
geological record in modern, non-lithifying microbial mats, this might not be true of Proterozoic
microbial mats that existed when carbonate precipitation was more favorable, thereby allowing
the upper layers to also have the possibility to lithify and enter the geological record.
Nonetheless, other studies have shown that precipitation of carbonate is correlated with the decay
of organic matter in the deeper mat layers and that the degree of calcification increases with
depth in the mat (Altermann et al., 2006), making our results here still applicable throughout the
geological record.
Another notable result is that the flat mat has the least amount of cyanobacterial filaments
(Fig. 9 and 10, Table A1). This is surprising because it would be expected that the old mat
samples, specifically the old flat mat, would have the least amount of cyanobacterial filaments
due to the fact that it has gone through more degradation and decay. Yet this not what we
observe. One explanation for this might be that the environmental conditions were different
when the old flat mat was at the surface. For example, perhaps the water depths were lower when
the old flat mat was at the surface, requiring it to adapt a strategy similar to the polygonal mats,
where it used empty cyanobacterial sheaths as a sort of ‘sunscreen’. In that case, the flat mat
would have been a polygonal mat when the old flat layer was at the surface. Indeed, changes in
the hydrology of Little Ambergris Cay have been documented after storm events (M. Gomes,
13
personal communication, April 2018). It is also possible that an environmental condition other
than water depth was different that either allowed for better cyanobacteria preservation, or for
there to be more cyanobacteria growth (e.g. nutrient levels). Testing these hypotheses would
require the age of the old flat strata to be determined, as well as the stated environmental
conditions at the time the old mat was at the surface.
As mentioned earlier, bacterial filament widths are one of the few properties that those
who study microfossils in the geological record have to identify and classify fossilized
microorganisms. Because Trembath-Reichert et al. (2016) found that cyanobacterial diversity
from 16S rRNA sequencing between the polygonal and flat mats on Little Ambergris Cay are
indistinguishable, we hypothesized here that we should obtain the same results using filament
widths. Over the filament width range of 0-5 μm, a range that most filament widths from the
mats occur in (Fig. 15-19 and Fig. A1), the filament widths between the polygonal mat and the
flat mat are significantly similar (Fig. 21). This result is consistent with the result of Trembath-
Reichert et al. (2016), and gives further credence to the idea that the differing morphologies of
the polygonal and flat mats is not due to the difference in their cyanobacteria populations.
Furthermore, the result from PCA that the polygonal mat pink layer and the flat mat pink layer
are more similar to the old flat mat and the old polygonal mat, respectively, rather than to the old
mat they lie directly above (Fig. 12d), is also evidence that the mats are all relatively similar in
terms of their bacterial filaments and microbial populations. Showing this using filament widths
rather than gene sequencing also allows us to say that filament widths can be used to observe
differences, or the lack thereof, between stromatolite morphotypes.
Over the entire range of filament widths observed in the mats, the filament widths are
significantly different between the polygonal and flat mats (Fig. 20). In particular, the polygonal
mat has filaments with widths greater than 5 μm that do not appear in the flat mat (Fig. A1).
Though this result contradicts the findings of Trembath-Reichert et al. (2016) that the
cyanobacterial populations are indistinguishable between the mat types based on 16S rRNA
analyses, it is possible that that study is capturing the similarity in the 0-5 μm range, which is
what is guiding their conclusions. Additionally, the fact that there are more thick cyanobacteria
filaments that have widths not observed in the flat mat could be consistent with the observation
of Trembath-Reichert et al. (2016) that the polygonal mat’s microbial population as a whole is
more diverse than the flat mat’s. When discussing the filaments in the 0-5 μm range, it is also
important to note that there are non-cyanobacterial filamentous bacteria with filament widths in
the 0-5 μm range that are being included in our filament width analyses. For example,
Chloroflexi are a phylum of filamentous bacteria that occur as thin filaments and are among the
most abundant taxa in both the polygonal and flat mats (Gomes et al. in revision; Trembath-
Reichert et al., 2016). Because we cannot distinguish between cyanobacterial and non-
cyanobacterial filaments based on morphology using SEM, it is unknown what part of the 0-5
μm range filaments Chloroflexi account for in this study. However, considering that the most
dominant cyanobacteria species occur as thin filaments, it is valid to assume that most filaments
in the 0-5 μm range actually are cyanobacteria.
Although the bacterial filaments that have widths ranging from 0-5 μm are
indistinguishable between the polygonal and flat mats, it is tempting to ask how cyanobacterial
diversity can be ruled out as a reason for the formation of the two morphotypes based on the
differences between their thick cyanobacterial filaments. However, there is a potential
environmental explanation for this observation that does not require an explanation based on
biological differences between the mats. The polygonal mats occur at a higher elevation than the
14
flat mats and therefore have more subaerial exposure (Gomes et al., in revision). The uppermost
portion of the mat is composed of empty cyanobacterial sheaths that contain the pigment
scytonemin, which acts as a protective layer against the harmful solar UV rays (Rastogi et al.,
2015; Balskus et al. 2011; Garcia-Pichel et al. 1992). The flat mat, which occurs at a lower
elevation and is therefore submerged, also has a protective layer, but it is made from EPS rather
than empty cyanobacterial sheaths. The polygonal mats cannot use EPS as a protective layer,
because it would dry out when the mats are subaerially exposed at low tide. Considering this, the
result that the polygonal and flat mats have differing filament widths above 5 μm (where the
ratio of filament widths greater than 5 μm to 0-5 μm is 0.61 for the polygonal mat and 0.077 for
the flat mat) and therefore possibly different cyanobacterial diversity can be explained by
environmental factors, meaning that environmental factors such as elevation shape mat
morphology rather than biology. The idea that the water depth the mats occur in affects their
morphology was put forth in Knoll et al. (2013). However, in that study they found that
stromatolite facies that occur in deeper water environments are more diverse than those that
occur in shallow waters. This is the reverse of what is observed at Little Ambergris Cay, where
the polygonal mats that occur at higher elevations and shallower water are more diverse than the
flat mats that are permanently submerged (Gomes et al., in revision; Trembath-Reichert et al.,
2016). Therefore, other environmental factors such as hydrodynamics and sedimentation likely
play a greater role in affecting the diversity of different mat morphotypes.
The filament width data analysis also shows that the old, buried mats are more similar to
the pink mat layers than the upper green layers. For the polygonal mat, the filament widths in the
old polygonal mat are similar to the polygonal pink layer, but not the polygonal green layer (Fig.
21). For the flat mat layers and the old flat mat, both the green and pink layers have significantly
different filament widths when compared to the old flat mat. However, over the 0-5 μm filament
width range, the pink layer has significantly similar widths to the old flat mat whereas the green
layer does not (Fig. 23). This means that filament widths do not change significantly as the pink
mat layers transition to the deeper brown layers, either because the filaments that remain in the
pink layer are more resistant to decay or because subsequent decay processes are not as effective
as those in the green layer. Further, because cyanobacterial and thin bacterial filaments are by far
the most abundant organisms in the mats, this trend is also observed when taking into account all
the organisms of the mats (i.e. from PCA and Bray-Curtis analysis). Taken together, these
observations indicate that filament widths measured in the geological record can be used to
extract paleoenvironmental information from lithified microbial mats.
In terms of the flat mat to polygonal mat succession hypothesis endorsed by Trembath-
Reichert et al. (2016) and others, there is not an entirely clear conclusion from this study. Each
metric or analysis used to characterize the mats relative to one another gives a different result.
From PCA, Bray-Curtis dissimilarity, and comparisons of 0-5 μm filament widths, it is clear that
the flat and old polygonal mats have similar properties. This may be an indication that the old
polygonal mat layer we sampled from could have once been a flat mat that developed after some
storm or other type of disruption. This interpretation would then be a validation of the flat to
polygonal succession hypothesis. On the other hand, the old flat mat shows similarity to the
polygonal mat pink layer in PCA (Fig. 12d), but also shows similarities to both the polygonal
and flat mats using the Bray-Curtis dissimilarity metric, though it is overall more similar to the
polygonal mat. Therefore, these results could just be another indication that the mats really are
similar in terms of the morphologies of their biological communities, and that environment is the
factor controlling macroscale mat morphologies. However, overall we have showed that the old
15
mats are not entirely the same as their active counterpartsan indication that these results either
lend validity to the mat succession hypothesis, or show that differing paleoenvironmental
conditions affected the abundance and lifestyle of mat microorganisms.
The final unexpected result was the observation of 70 μm wide filaments in the deep
layers of the polygonal mat (Fig. 17). Thick filaments have been observed in the geological
record, for example in microbialites from the Angmaat Formation that contain filaments ascribed
to the extant mat-building cyanobacteria Microcoleus chthonoplastes (Knoll et al., 2013).
However, this observation of thick filaments in the deep layer laid down by previous generations
of mat communities is surprising because filaments this wide were not observed in the surface
layers of the polygonal mat. Two possibilities may account for this result. First, there may have
been environmental conditions at the time the old polygonal mat was at the surface that were
more advantageous to large (70 μm wide) filamentous cyanobacteria. In this scenario,
environmental conditions were different when the new, active mat was collected such that they
did not allow for the growth of cyanobacteria with large filament widths. The next possibility is
that as the mats were buried and organic matter from decaying organisms began to accumulate,
cyanobacterial filaments were covered in this material, thereby amplifying the widths we
observe. This is a plausible explanation given the appearance of the cyanobacteria filaments in
the active polygonal mat, which do not have much material building up on them, and the
filaments from the old polygonal mat, which appear to have material that has built up over time
(Fig. 25). This notion also goes back to the conclusions from Newman et al. (2017), who found
that sheathed cyanobacteria can accumulate mineral veneers of up to 1 μm in thickness around
their filaments. Although the Little Ambergris mats are non-lithifying, and therefore
cyanobacteria filaments likely do not accumulate mineral veneers, it is not unreasonable to think
that the same idea may apply to organic matter. However, 1 μm accumulation does not explain
the >30 μm increase in filament widths, or the fact that this increase does not apply to the whole
range of filament widths, which is what we would expect. Additionally, not all thick
cyanobacterial filaments in the old mats were observed to have accumulated material, so it is still
unclear how this can affect filament widths observed in these mats and the geological record.
Therefore, the former hypothesis of environmental change causing different organisms to grow is
the more likely explanation.
Of course, there are limitations to this study that should be addressed. First, we only have
samples from two different mats from two locations, so it isn’t clear what variability may exist
using the methods outlined here between different mats of the same morphotype. However, the
microbial diversity results indicate that the surface polygonal (flat) mat are more similar to other
polygonal (flat) mats than they are to the flat (polygonal) mats, based on similarities between the
surface mats and the old mats. Also, though SEM is useful for looking at the overall structure of
the mats, other types of microscopy such as light microscopy may be able to yield
complementary, or more information on the organisms present in the mats. Future work therefore
may include obtaining more mat samples for SEM imaging, or resampling the same mats from
this study and using different methods and techniques to analyze them. Another limitation of this
study is that when considering the differences between the active mats and the buried mats, we
did not take into account the fact that differences between them may arise because of the decay
and loss of morphological signatures. Future work can include an assessment of this based on the
results of taphonomic studies of Little Ambergris Cay mats such as Gomes et al. (in revision).
Future work will also elucidate how the methods used in this study can be used in tandem with
other methods available to geologists studying lithified microbial mats, such as stable isotope
16
analyses, to answer lingering questions. Such questions include, for instance, can lithified
microbial mats record global environmental change, or only localized changes in hydrology and
sedimentation? Answering these questions will enable researchers to better understand how
microscopic methods can be employed to study lithified microbial mats in the geological record,
and can also shed new light on data obtained in older studies. Finally, there are significant
environmental and biological differences between the present day when Little Ambergris Cay
mats exist and the Precambrian, when the stromatolites that we seek to better understand existed.
These differences place some limits on what we can infer about stromatolites based on what we
observe in modern microbial mats. For example, atmospheric oxygen concentrations would have
been lower than they are today, and diatoms and other eukaryotic taxa would not have evolved
yet, changing the ecosystem of the mats. However, the abundance of cyanobacteria microfossils
in stromatolites and the similar shallow, coastal marine environments these ancient microbial
mats occur in shows that in general, the most salient conditions were similar.
Overall, from this study we can see that paleoenvironmental conditions have the potential
to be extrapolated on the basis of morphology by observing the differences between different
microbial mat morphotypes. These results from modern microbial mats can be extrapolated back
to the geological record and can ultimately serve to help us better interpret stromatolites and
understand environmental change throughout Earth history.
5. Conclusions
In the past, the study and interpretation of lithified microbial mats in the geological
record has been made difficult by the destruction and diagenetic overprinting of primary features.
This has made the observation that stromatolite morphologies and textures change through time
difficult to interpret. Recent studies of modern microbial mats with varying morphologies have
helped to make steps towards resolving this issue, because we have tools such as gene
sequencing to determine biodiversity and can directly correlate biodiversity and environmental
factors with mat morphologies. In this study, we used observations of two modern microbial mat
morphotypes from Little Ambergris Cay obtained from the tools that are unavailable for studying
the geological record (e.g. gene sequencing) and compared those observations to the ones
obtained here through microscopy, which is used by researchers studying stromatolites, to see if
similar observations can be made in the geological record. By sampling from the old, buried
mats, we also tried to determine if paleoenvironmental information can be obtained from these
records.
Through these indirect methods we have found that the polygonal and flat mats differ in
terms of their microbial assemblages and that cyanobacterial filament widths in the 0-5 μm range
are indistinguishable between the two mat morphotypes, findings that are consistent with studies
of the same mat types using 16S rRNA gene sequencing. Additionally, we found that
cyanobacterial filament widths and microfauna assemblages in the pink layers of the active mats
are more similar to the old mats than the green layers are. These findings are consistent with
results that were obtained using organic geochemistry methods, which indicate that the greatest
rate of degradation occurs in the green layer, and that subsequent degradation in the pink layer
and lower layers is reduced.
We also found that the surface mats do not show relative similarity to their buried
counterparts, suggesting that we are capturing the effects of a lack of preservation of some taxa,
or varying environmental conditions through time. In addition, the surface mats are more similar
to the buried counterpart of the other mat type, e.g. the flat mat and old polygonal mat are more
17
similar than the flat mat and old flat mat. Moreover, the active flat mat has less thick
cyanobacterial filaments than the old flat mat, which is a possible indication that environmental
conditions were different when the old flat mat was at the surface, and perhaps was a polygonal
mat that needed to adapt differently than a flat mat. All of these observations are potential
evidence for the mat succession hypothesis and/or evidence for differing environmental
conditions that required different adaptations.
Though we obtained conclusive and consistent results from this study, there is still some
uncertainty and gaps in our knowledge. For instance, we still need to determine if the results
from this study are repeatable, and how applicable they are across other microbial mat
morphotype and environment systems. Overall, based on the findings in this study, the varying
macroscale morphology of microbial mats from Little Ambergris Cay is not due to differences in
the populations of cyanobacteria, the primary mat builders. The differing morphology of these
mats through space and time is explained by the differing environmental conditions. Because
these findings are consistent with those found using the tools employed by those studying
modern systems, certain paleoenvironmental information can be obtained from lithified
microbial mats using microscopy. However, considerations of the macroscale morphology of
lithified microbial mats is also important, as these can be more easily observed and contribute
complementary knowledge. Finally, this study is a further indication that lithified microbial mats
are important records of not only the appearance of cyanobacteria and other taxa, but of
environmental changes and patterns throughout Earth history.
Acknowledgements
Thank you to Dr. Maya Gomes for her help in all aspects of this project and for her
guidance and unwavering support throughout. Thanks also to Dr. Michael McCaffrey from the
IIC for teaching me all about scanning electron microscopy and for his helpful insight and
suggestions that made this project possible. Thanks to Dr. Scot Miller and Dr. Meghan Avolio
for their help with the statistical methods. Finally, thank you to the JHU Earth and Planetary
Sciences department for giving me this opportunity.
References
Altermann, W., Kazmierczak, J., Oren, A. and Wright, D.T., 2006. Cyanobacterial calcification
and its rockbuilding potential during 3.5 billion years of Earth history. Geobiology, 4(3),
pp.147-166.
Andres, M.S. and Reid, R.P., 2006. Growth morphologies of modern marine stromatolites: a case
study from Highborne Cay, Bahamas. Sedimentary Geology, 185(3-4), pp.319-328.
Balskus, E.P., Case, R.J. and Walsh, C.T., 2011. The biosynthesis of cyanobacterial
sunscreen scytonemin in intertidal microbial mat communities. FEMS microbiology
ecology, 77(2), pp.322-332.
Barghoorn, E.S. and Tyler, S.A., 1965. Microorganisms from the Gunflint chert. Science,
147(3658), pp.563-577.
Benzerara, K., Skouri-Panet, F., Li, J., Férard, C., Gugger, M., Laurent, T., Couradeau, E.,
Ragon, M., Cosmidis, J., Menguy, N. and Margaret-Oliver, I., 2014. Intracellular Ca-
carbonate biomineralization is widespread in cyanobacteria. Proceedings of the National
Academy of Sciences, 111(30), pp.10933-10938.
18
Boal, D. and Ng, R., 2010. Shape analysis of filamentous Precambrian microfossils and modern
cyanobacteria. Paleobiology, 36(4), pp.555-572.
Des Marais, D.J., 2003. Biogeochemistry of hypersaline microbial mats illustrates the dynamics
of modern microbial ecosystems and the early evolution of the biosphere. The Biological
Bulletin, 204(2), pp.160-167.
GarciaPichel, F., Sherry, N.D. and Castenholz, R.W., 1992. Evidence for an Ultraviolet
Sunscreen Role of the Extracellular Pigment Scytonemin in the Terrestrial
Cyanobacterium Chiorogloeopsis sp. Photochemistry and Photobiology, 56(1), pp.17-23.
Gomes, M. et al., 2017. Microbial Mats on Little Ambergris Cay, Turks and Caicos Islands:
Taphonomy and the Selective Preservation of Biosignatures. Accepted to Geobiology.
Grotzinger, J.P. and Rothman, D.H., 1996. An abiotic model for stromatolite morphogenesis.
Nature, 383(6599), p.423.
Grotzinger, J.P. and Knoll, A.H., 1999. Stromatolites in Precambrian carbonates: evolutionary
mileposts or environmental dipsticks?. Annual review of earth and planetary sciences,
27(1), pp.313-358.
Knoll, A.H., Wörndle, S. and Kah, L.C., 2013. Covariance of microfossil assemblages and
microbialite textures across an upper Mesoproterozoic carbonate platform. Palaios,
28(7), pp.453-470.
Mackey, T.J., Sumner, D.Y., Hawes, I., Jungblut, A.D. and Andersen, D.T., 2015. Growth of
modern branched columnar stromatolites in Lake Joyce, Antarctica. Geobiology, 13(4),
pp.373-390.
Newman, S.A., KlepacCeraj, V., Mariotti, G., Pruss, S.B., Watson, N. and Bosak, T., 2017.
Experimental fossilization of matforming cyanobacteria in coarsegrained siliciclastic
sediments. Geobiology, 15(4), pp.484-498.
Orzechowski, E. A., Strauss, J. V., Knoll, A. H., Fischer, W. W., Cantine, M., Metcalfe, K.,
Quinn, D. P., Stein, N., Gomes, M. L., Grotzinger, H. M., Lingappa, U., O'Reilly, S. S.,
Riedman, L. A., Trower, L., and Grotzinger, J. P., Age and Construction of Little
Ambergris Cay Bedrock Rim, Southeastern Caicos Platform, British West Indies, in
Proceedings American Geophysical Union, Fall General Assembly 2016.
Peters, S.E., Husson, J.M. and Wilcots, J., 2017. The rise and fall of stromatolites in shallow
marine environments. Geology, 45(6), pp.487-490.
Rastogi, R.P., Sonani, R.R. and Madamwar, D., 2015. Cyanobacterial sunscreen scytonemin:
role in photoprotection and biomedical research. Applied biochemistry and
biotechnology, 176(6), pp.1551-1563.
Riding, R., 1992. Temporal variation in calcification in marine cyanobacteria. Journal of the
Geological Society, 149(6), pp.979-989.
Ruxton, G.D., 2006. The unequal variance t-test is an underused alternative to Student's t-test
and the MannWhitney U test. Behavioral Ecology, 17(4), pp.688-690.
Schopf, J.W., 1993. Microfossils of the Early Archean Apex chert: new evidence of the antiquity
of life. Science, 260(5108), pp.640-646.
Shixing, Z. and Huineng, C., 1992. Characteristics of Palaeoproterozoic stromatolites in China.
Precambrian Research, 57(1-2), pp.135-163.
Stein, N., Grotzinger, J. P., Quinn, D. P., Fischer, W. W., Knoll, A. H., Cantine, M., Gomes,
M.L., Grotzinger, H. M., Lingappa, U., Metcalfe, K., O'Reilly, S. S., Orzechowski, E. A.,
Riedman, 1235 L. A., Strauss, J. V., and Trower, L., UAV, DGPS, and laser transit
19
mapping of microbial mat ecosystems on Little Ambergris Cay, B.W.I in Proceedings
American Geophysical Union, Fall General Assembly 2016.
Trembath-Reichert, E., Ward, L.M., Slotznick, S.P., Bachtel, S.L., Kerans, C., Grotzinger, J.P.
and Fischer, W.W., 2016. Gene Sequencing-Based Analysis of Microbial-Mat
Morphotypes, Caicos Platform, British West Indies. Journal of Sedimentary Research,
86(6), pp.629-636.
Wanless, H.R., Tyrell, K.M, Tedesco, L.P., and Dravis, J.J., 1988. Tidal-flat
sedimentation from Hurricane Kate, Caicos Platform, British West Indies: Journal of
Sedimentary Petrology, v. 58, p. 724738.
20
Figures
Figure 1. Violin plot” from Stein et al. (2016) showing that the EPS-coated flat mats occur at the
lowest elevations, the polygonal mats at intermediate elevations, and the blister mats at the highest
elevation. Low elevation correlates with greater water depth and high elevation correlates with
shallower water.
21
Figure 2. Location of Little Ambergris Cay within the Caicos Platform (left) and location of Turks
and Caicos within the southwest north Atlantic Ocean (left, inset). The location of microbial mat
study sites on Little Ambergris Cay (right) show where the polygonal mat (CC17-4) and flat mat
(CC3) were sampled from. Figure used with permission from M. Gomes.
22
Figure 3. Images of the flat and polygonal microbial mats from Little Ambergris Cay. Top left:
EPS-covered flat mats; Bottom left: vertical cross section through CC3 showing the green, pink,
and old brown layers; Top right: tufted polygonal mats; Bottom right: vertical cross section
through CC17-4 showing the green, pink, and old brown layers; Photo credit: Maya Gomes.
23
Figure 4. Cross section of polygonal mat CC17-4 showing the notable mat layers: G-green, P-
pink, O1-ooid layer 1, O2-ooid layer 2, OP-old polygonal mat. Note the filamentous texture of OP.
Layers G, P, and OP are the subjects of this study.
G
P
O1
O2
OP
24
Figure 5. Cross section of flat mat CC3 showing the notable mat layers: G-green, P-pink, OF-old
flat mat. The top orange-tan layer is composed of granular EPS. Also note the granular texture of
OF. Layers G, P, and OF are the subjects of this study. Photo credit: Maya Gomes.
G
P
25
Figure 6. Representative images of the polygonal and flat mats, and their respective layers. From
left to right: green layer of the polygonal mat, pink layer of the polygonal mat, old polygonal mat,
green layer of the flat mat, pink layer of the flat mat, old flat mat.
26
Figure 7. Mat features expressed as percent relative abundance from 100-point counts across
twenty sites in the polygonal, flat, old polygonal, and old flat mats. Bars for features with relative
abundance <5% are too small to appear. Data in Table A1.
Figure 8. Mat features from the individual mat layers expressed as percent relative abundance
from 100-point counts across twenty sites in the polygonal, flat, old polygonal, and old flat mats.
Bars for features with relative abundance <5% are too small to appear. Data in Table A1.
0 102030405060708090 100
Relative abundance (%)
Old flat
Flat
Old polygonal
Polygonal
Amorphous organic matter
Cyanobacterial filaments
Thin bacterial filaments
Coccoidal bacteria
Colonial coccoids
Diatoms
Eukaryote feces
Dinoflagellates
Larval mollusk
0 102030405060708090 100
Relative abundance (%)
Polygonal green
Polygonal pink
Old polygonal
Flat green
Flat pink
Old flat
Amorphous organic matter
Cyanobacterial filaments
Thin bacterial filaments
Coccoidal bacteria
Colonial coccoids
Diatoms
Eukaryote feces
Dinoflagellates
Larval mollusk
27
Figure 9. Mat features (without amorphous organic matter) expressed as percent relative
abundance from 100-point counts across twenty sites in the polygonal, flat, old polygonal, and old
flat mats. Bars for features with relative abundance <5% are too small to appear. Data in Table
A1.
Figure 10. Mat features (without amorphous organic matter) expressed as percent relative
abundance from 100-point counts across twenty sites in the polygonal, flat, old polygonal, and old
flat mats. Bars for features with relative abundance <5% are too small to appear. Data in Table
A1.
0 102030405060708090 100
Relative abundance (%)
Old flat
Flat
Old polygonal
Polygonal
Cyanobacterial filaments Thin bacterial filaments Coccoidal bacteria Colonial coccoids Diatoms Eukaryote feces Dinoflagellates Larval mollusk
0 102030405060708090 100
Relative abundance (%)
Old flat
Flat pink
Flat green
Old polygonal
Polygonal pink
Polygonal green
Cyanobacterial filaments Thin bacterial filaments Coccoidal bacteria Colonial coccoids Diatoms Eukaryote feces Dinoflagellates Larval mollusk
28
Figure 11. PCA plots of point count data for bulk mats and individual mat layers. Points represent
relative abundance data: a,b) with amorphous organic matter, c,d) normalized without amorphous
organic matter. Principal components 1 and 2 explain: a) 84% b) 72% c) 85% and d) 68% of the
variance.
a.
))
)
b.
))
)
c.
))
)
d.
))
)
29
Figure 12. PCA plots of point count data for bulk mats and individual mat layers. Points represent
relative abundance data: a,b) with amorphous organic matter, excluding singletons, c,d) without
amorphous organic matter, excluding singletons. Principal components 1 and 2 explain: a) 93% b)
88% c) 97% and d) 93% of the variance.
a.
))
)
b.
))
)
c.
))
)
d.
))
)
30
Figure 13. Bray-Curtis similarity matrix for the bulk (green and pink layers) mats. Color scheme
indicates percent similarity. P=polygonal mat, F=flat, OP=old polygonal, OF=old flat.
Figure 14. Bray-Curtis similarity matrix for the individual mat layers. PG=polygonal mat green
layer, PP=polygonal pink, OP=old polygonal, FG=flat green, FP=flat pink, OF=old flat.
P F OP OF
P
F
OP
OF 89.3
90.4
80.5
90.4
86.6
89.3
80.5
86.6
70.9
76.4
70.9 76.4
72
74
76
78
80
82
84
86
88
90
NaN
Similarity
(%)
PG PP OP FG FP OF
PG
PP
OP
FG
FP
OF
70.7
75
70.7
89.3
74.9
83.7
92.7
89.3
80.9
87.9
74.9
80.9
80.4
70.9
83.7
87.9
80.4
80.4
75
92.7
70.9
80.4
62.1
56.1
55.9
62.1 56.1 55.9
60
65
70
75
80
85
90
NaN
Similarity
(%)
31
Figure 15. Histogram of filament widths from across all four mat samples. Widths were obtained
from bacterial filaments identified during 100-point counts across twenty sites in the polygonal,
flat, old polygonal, and old flat mats. Bars for widths greater than 45 micrometers are too small
appear in the plot.
32
Figure 16. Histograms of filament widths from the green and pink layers of the polygonal mat,
and the total filament widths from both layers (bottom panel). Statistical data in Tables A2 and
A3.
Figure 17. Histogram of filament widths from the old polygonal mat. Statistical data in Tables A2
and A3.
33
Figure 18. Histograms of filament widths from the green and pink layers of the flat mat, and the
total filament widths from both layers (bottom panel). Statistical data in Tables A2 and A3.
Figure 19. Histogram of filament widths from the old flat mat. Statistical data in Tables A2 and
A3.
34
Figure 20. p-value matrix for comparisons of bacterial filament across the entire filament width
range of the mats. Higher p-values have darker colors. Significance level is 0.05. P=polygonal,
F=flat, OP=old polygonal, OF=old flat. Corresponding T-statistics in Table A4.
Figure 21. p-value matrix for comparisons of bacterial filament across the entire filament width
range of the mats. Higher p-values have darker colors. Significance level is 0.05. PG=polygonal
mat green layer, PP=polygonal pink layer, OP=old polygonal, FG= flat green, FP=flat pink,
OF=old flat. Corresponding T-statistics in Table A4.
P F OP OF
P
F
OP
OF
3.4e-09
0.08
3.4e-09
0.001
2.6e-06
0.08
0.001
0.3
2.6e-06
0.3
0.5
0.5
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0.5
NaN
PG PP OP FG FP OF
PG
PP
OP
FG
FP
OF
0.3
0.03
3.6e-06
1.1e-06
0.3
0.3
0.3
0.0002
0.0001
0.03
0.3
0.004
0.003
0.3
3.6e-06
0.0002
0.004
6.3e-05
1.1e-06
0.0001
0.003
1.8e-05
0.3
0.3
6.3e-05
1.8e-05
1
0.9
0.9
1
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
NaN
35
Figure 22. p-value matrix for comparisons of bacterial filament across the filament width range
0-5 μm, where most filament widths occurred. Higher p-values have darker colors. Significance
level is 0.05. P=polygonal, F=flat, OP=old polygonal, OF=old flat. Corresponding T-statistics in
Table A5.
Figure 23. p-value matrix for comparisons of bacterial filament across the filament width range
0-5 μm, where most filament widths occurred. Higher p-values have darker colors. Significance
level is 0.05. PG=polygonal mat green layer, PP=polygonal pink layer, OP=old polygonal, FG=
flat green, FP=flat pink, OF=old flat. Corresponding T-statistics in Table A5.
P F OP OF
P
F
OP
OF
0.1
0.05
0.1
0.06
0.1
0.1
0.001
0.05
0.06
0.001
0.9
0.9
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
NaN
PG PP OP FG FP OF
PG
PP
OP
FG
FP
OF
0.03
0.004
0.07
0.4
0.03
0.6
0.2
0.01
0.004
0.6
0.6
0.5
0.001
0.07
0.6
0.3
0.04
0.4
0.2
0.5
0.3
0.3
0.01
0.001
0.04
0.3
0.8
1
1
0.8
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
NaN
36
Figure 24. Depth profiles showing the degradation and depletion of labile organic molecules
through the green, pink, and brown layers of a polygonal mat from Gomes et al. (in revision).
Figure used with permission from Gomes et al. (in revision).
37
Figure 25. Left: Cyanobacterial filaments from the green layer of the polygonal mat that are not
covered in much amorphous organic matter. Right: Thick cyanobacterial filament from the old
polygonal mat that is completely covered in amorphous organic matter.
38
Appendix
Table A1. Absolute abundances of mat features and taxa from point counting.
Feature/Taxa
Polygonal
green
Polygonal
pink
Flat
green
Flat pink
Old
polygonal
Old flat
Amorphous
organic matter
1343
2118
1856
1851
3526
3790
Cyanobacterial
filament
109
45
5
7
30
48
Bacterial
filament
103
151
63
93
151
112
Coccoid
27
68
54
41
90
46
Colonial
Coccoid
16
10
2
0
0
2
Diatom
1
3
18
8
1
1
Eukaryote
feces
0
4
1
0
0
0
Dinoflagellate
0
0
0
0
1
0
Larval mollusk
0
0
0
0
0
1
Table A2. Filament width statistics for mats and their respective layers over the whole filament
width range. P=polygonal, F=flat, PG=polygonal green, PP=polygonal pink FG=flat green,
FP=flat pink, OP=old polygonal, OF=old flat. Mean widths are reported in micrometers and n is
number of filaments measured.
P
PG
PP
F
FG
FP
OP
OF
n
289
121
168
172
76
96
195
158
Mean
7.9
8.6
7.4
3.3
3.2
3.3
6.2
7.3
Standard
deviation
10.6
9.8
11.2
5.8
6.1
5.5
10.4
8.9
Variance
113.0
95.3
125.1
33.3
37.5
30.0
107.1
79.8
Table A3. Filament width statistics for mats and their respective layers for the filament width
range 0-5 µm. P=polygonal, F=flat, PG=polygonal green, PP=polygonal pink FG=flat green,
FP=flat pink, OP=old polygonal, OF=old flat. Mean widths are reported in micrometers and n is
number of filaments measured.
P
PG
PP
F
FG
FP
OP
OF
n
214
80
134
159
71
88
158
110
Mean
2.0
1.8
2.1
2.0
2.1
2.0
2.2
1.8
Standard
deviation
1.0
0.8
1.1
1.0
1.1
0.9
1.1
0.8
Variance
1.0
0.6
1.1
0.9
1.2
0.8
1.2
0.7
39
Table A4. T-statistics from Welch’s t-test. Tests cover whole filament width range.
P=polygonal, F=flat, PG=polygonal green, PP=polygonal pink, FG=flat green, FP=flat pink,
OP=old polygonal, OF=old flat.
Comparison
T-statistic
P-F
6.0
OP-P
-1.8
OF-P
-0.6
OP-F
3.3
OF-F
4.8
OP-OF
1.1
PG-PP
1.0
PG-FG
4.8
PG-FP
5.0
PP-FP
3.9
PP-FG
3.7
FG-FP
-0.15
OF-FG
4.1
OF-FP
4.4
OF-PG
1.2
OF-PP
-0.06
OP-FG
2.9
OP-FP
3.0
OP-PG
-2.1
OP-PP
-1.1
Table A5. T-statistics from Welch’s t-test. Tests cover filament width range 0-5 µm.
P=polygonal, F=flat, PG=polygonal green, PP=polygonal pink, FG=flat green, FP=flat pink,
OP=old polygonal, OF=old flat.
Comparison
T-statistic
P-F
-0.08
OP-P
-1.7
OF-P
2.0
OP-F
-1.5
OF-F
1.9
OP-OF
3.3
PG-PP
-2.3
PG-FG
-1.8
PG-FP
-0.9
PP-FP
1.4
PP-FG
0.02
FG-FP
1.1
OF-FG
-2.1
OF-FP
-1.1
OF-PG
-0.2
OF-PP
-2.6
OP-FG
0.5
40
OP-FP
2.0
OP-PG
2.9
OP-PP
0.6
Figure A1. Histograms of filament widths from all mat types and layers. The bin number (70)
was chosen to show the filament width distributions at a higher resolution than those in Figures
15-19.